Bell Canada → [Internet] Random one minute connection drops - Bell Fibe 1.5

Was PingPlotter on wired or wireless PC?

It was connected via Ethernet to the 10gb port on the Gigahub. Wifi on the MacBook was disabled.

If both your Bell and Rogers lines are going down, there's something else going wrong, likely with your configuration or hardware.

What's your network layout like?

Have you verified the physical layer?

Once you've done that...

Have you verified the physical layer again?

I had a client who's dog had nicked a section of their Ethernet cable, and it resulted in intermittent high packet loss.

In my teenage years, I personally had an issue where an Ethernet cable that ran along the ground in a high traffic part of the house started to have issues after years of being stepped on occasionally, causing high packet loss intermittently.

I know it sounds stupid, but the first and second rule of networking is check the physical layer for a reason. This was drilled into us at school and has proved to be right more times than not.

No, the OP meant pings to Bell and Rogers DNS servers, not two different lines. It's good to read the thread in full before replying.

I have.
I misunderstood the pinging of Rogers DNS servers, because, in general pinging DNS servers is not a good idea because they don't prioritize responding to pings, and ISPs DNS servers are notoriously unreliable. Normally you wouldn't use them unless you needed too, and I'm actually surprised Rogers' DNS actually replies to queries outside of their network. It's common practice to make DNS servers only reply to requests from clients on it's local network. This is why I thought they had a Rogers connection as well.

Common? Maybe 20 years ago. Can you post one example of some ISP DNS server that doesn't respond outside pings or requests? I have pinged some small ISP DNS server for a few years as part my gateway monitoring and it's still responding.

I didn't say ISP DNS servers. I said DNS servers. Like the internal DNS servers at Apple while I worked there. Apple, since they own a /8, assigns Class A IP addresses to all their internal network equipment, even those firewalled off from the outside world. Their DNS servers being an example. This is also the case for a lot of corporate networks, where even though their DNS servers might be publicly addressable, they will not respond to queries outside of their IP space.

Granted this is a more meaningless statement now that most companies are using NAT and private IP spaces for their internal network, but the point still stands: I did not expect Rogers' equipment to reply to requests off their network, and pinging a DNS server of any kind is generally a bad idea because DNS servers get lots of requests and de-prioritize things like ICMP requests.

Why did you even say it? If it's a server doesn't respond to queries outside of their IP space then nobody would be able to ping it. If it responds it will continue to respond. The world is changing every day and those who adapt thrive.

I don't know why you've chosen to become so combative with me, but you could simply not read or reply to my comments if they bother you so much.

The reason I brought it up was as an explanation for why I thought they also had a Rogers connection.

Hey I appreciate both of you helping me attempt to diagnose these connection issues.

Can we please return the thread to that discussion?

Could the frequency of these disconnections (every 20-30 mins) suggest an issue with the account configuration in the Gigahub?

Given the data I now have that shows the disconnections happening not only on wifi, would you suggest I go back to Bell with the results?

How best can I characterize these results to inform Bell tech support and get them to actually reconfigure things on their end?

Yep, for sure. Sorry about that.It really depends on where the disconnections are happening, like what part of the chain.That's gonna be the tough part...

Here's what I would do next: Run pings between your computer and the GigaHub to see if your connection between your computer and the GigaHub are getting interrupted.

Is it safe to assume that you're directly wired to the GigaHub? (I recall you saying this, but I can't find the comment right now) If you are, great, if not, we'll need to debug every hop in the chain.

If you can catch it happening, and try to ping from the GigaHub directly to the outside world, while it's happening then that would help you further.

Yes I was connected directly to the Gigahub during the testing.

How would you ping from the Gigahub to the outside world exactly?

This may indicate a routing problem. Above that plot in PingPlotter there is a trace route showing hops through various routers on the internet. When you double-click on the graph at a particular time on your lower graph (where the red is) the upper part will update. Can you see where the red (packet loss) starts on the upper graph? The first hop is your router/GigaHub, the second hop is Bell's gateway and so on. Look for the hop where red starts and continues to the end. What is it? Some routers don't respond to ping, so you may have "red" hops and then "green", those don't count.

Please guys, keep the personalities out of threads. As soon as you say "you ...." things start to get personal and can become accusatory. There's interesting info in this thread so I'm not deleting stuff ... but parts came pretty close.

Thanks!

In one of the buttons on the right of the Home Hub/GigaHub interface there's a button called something along the lines of "advanced tools", I believe the ping option is in there. That should let you ping the outside world directly from your GigaHub.

You can also start running a ping to your Gigahub directly from your computer using PingPlotter too, which assuming it's at it's default IP address, it should be 192.168.2.1. See if you have issues with that at the same time you have issues communicating with the outside world.

ok thanks, found this in the options and will try it out the next time I'm testing

I've never used Ping Plotter before, but something is standing out to me in that first photo. It looks like Ping Plotter is saying that it couldn't even reach your Gigahub during that last outage... again if I'm reading it properly. Never having used the tool, I could be misunderstanding...

But that would indicate that there's either an issue with the Gigahub, or your link to the Gigahub.

Add a ping to 192.168.2.1 and see if that goes down with the rest of your internet when this happens.

I concur. Your traffic stops at the GigaHub. What's interesting though, the latency increases step-wise before 100% packet loss happens. Another "feature" I noticed, in your third graph the minimum ping time to your gateway (the second hop, 142.124.37.182) is very high compared to the other hops (134 vs. 7 and 28 ms). I wonder if something happens at the gateway before you lose your connection. But you would have something in the GigaHub log at those times, either some "O" errors or PPPoE renegotiations. Can you double check the GigaHub log and make sure its system time corresponds to your PC's?

I know you said you were connected directly to the GigaHub 10Gb port but can you confirm that there is absolutely no switch between your PC and the GigaHub?
It could also be a bad cable.
Also, can you try another port (1 Gb) on the GigaHub?
Do wireless outages happen at exactly the same times as wired? If so, my previous questions about the switch, port and cable are probably irrelevant.

You mentioned some account configuration issues. I'd recommend changing your b1 password. It's not your online account password but the one used for PPPoE sessions. You can do it from your online account portal.
I would also recommend factory resetting your GigaHub. This procedure often helps resolve weird issues with the GigaHub based on older posts here. Change the password online, wait for about 5 min and factory reset the GigaHub. It should automatically pick up the new password on first boot. You will need to re-setup the GigaHub if you made any changes in it. Changing the password may (or may not) help clear previous configuration issues.

Before you do the above can you post what ping times you have when everything is OK. Double click on that flat line in PingPlotter when nothing is going on.

If you get 100% packet loss to the GigaHub something is not right about it. Can you double check it's not rebooting at the times when the outages happen? It should be in the logs. And if you get a chance to be next to it when this is happening you will see something on its display. But see if password change and factory reset help.

If your GigaHub wouldn't give you 100% packet loss when an outage happens I would blame your gateway (the second hop in trace route). But your GigaHub kind of dies and stops responding to ping. I know you tired several GigaHubs but can you find another power supply to test? If your GigaHub stops responding to ping all hops beyond it would show 100% packet loos too.

I wanted to blame your gateway (and thus your local Bell network) because in many graphs you posted the maximum and average latencies for it are high, higher than other hops. But I don't know how to explain your GigaHub not responding to ping.

Can you post one more graph and focus on that step-wise increase of latency that is prominent on the first graph of your latest post? You also have other parts of the graph where you have increased latency but that doesn't lead to outages.

You can probably also post this on the Bell community forum and refer them to this thread. The things are not easy to explain.

Don't want to count my chickens quite yet but I haven't experienced an outage since changing the b1 password!

Also factory reset the Gigahub afterwards but that was attempted many times before and it didn't seem to resolve the issue.

Unfortunately my trial period for PingPlotter has run out and it's too expensive to subscribe for me. Know of any other tools I could use to do some more testing?

In any case, I'll report back in a week if I don't have any outages. Fingers crossed!

Thanks again for all the help!

Thanks for the update! Sounds promising. Can you install PingPlotter on another PC as trial? Or maybe someone knows of other similar free tools.

Keep us posted.

Depending on your level of expertise, you can set up a Linux server with Telegraph InfluxDB and Grafana to capture and graph data in a similar way to PingPlotter.

That's what I've done for my home network.

However these tools are all geared towards IT professionals, so you have to be comfortable around a Linux command terminal to be able to set these up.